Designs Of Sensor Matrixes For Biomimetic Electronic Nose And Their Application In Detection Of Beef Quality

Nowadays the output of meat increases quickly. And the problems of security of meat products got more and more attention because of a few toxic events about meat.The conventoinal methods of examining meat characters include subjective-judging and chemical methods. The subjective-judging methods depend on the sense (such as scent, seeing and taste) and experience of persons who make estimatation to a large degree. So the persons' physiological state will influce the results of judgement which is the reason of short of reproducibility in subjective judging. The chemical methods include gas chromatography, high performance liquid chromatography, atomic spectrophotometry, hydride generation -atomic fluorescence spectrometry , high performance capillary electrophoresis and so on. And these methods have the characters of excellent nicety and reproducibility compared with subjective-judging methods. But there are some shortcoming inluding high requirement of sample preparation and detection conditions in chemical methods. So these time-consuming methods can not satisfy the requirement of market.Bionics is developmening vigorously and applied in a lot of fields such as architecture, material, configuration and framework and so on. This study focuses on the study of bionic application to electronic nose system which can be used in checking beef quickly. The gas sensor and artificial NN arithmetic have been used in developing this kind of electronic nose.There are three main parts in bionic electronic nose system: gas sensor array and nose channel system; circuit system; software analysis system of computer. The main function of gas sensor array and nose channel system is to react to odor which the sample emanates. The gas sensor can translate that reaction into output of voltage signals. The sensor array is constituted with 6 gas sensors, all gas sensors are bought from shop, so they are cheap and reliable. The circuit can process low voltage signals come from the sensor, it is composed by filtering, amplifying processing, translating analog signals into digital signals and delivery it to computer. The part of software processes the digital signals, including of storage, describing curve, processing with artificial neural network algorithm and drawing conclusions.Studying and summarizing the biologic characteristic is the key of bionics. The difference between human olfaction system and dogs' olfaction system were compared. Working mode of their olfaction systems are the same ones: olfactory response induced by the inhaled gas molecular can be passed to the cerebra to be judged. The main reason why dogs' olfaction is acuter than human olfaction lies in difference between olfaction cellulars and configutation of nasal cavities. The vantage point of this research lies in configuration bionics. Dog s' olfaction cells distribute in the middle of nose channel which make olfaction cells can contact more odor molecules and make olfaction more acuter.Based on the bionic analysis and biologic configuration characteristic, the system of nose channel has been devised in this study. The nose channel system has thin air inlet and thick round piping outlet. The gas sensor array lie in the middle of separator on which some gas-guide holes can lead airflow and make more odorant molecules contact with gas sensors. An induced draft fan on air outlet can adjust the velocity of gas which can make backset in nose channel. It also can make more odorant molecules contact with gas sensors.A single gas sensor has poor selectivity and cross sensitivity to many kinds of gas in different degreee. If the gas sensor array could be established, the error of single gas sensor can be reduced. The efficient processing method of multi-signals amalgamation is essential in applying gas sensor array. And the new-emerging artificial neural network system has inimitable predominance. Artificial neural network is a kind of pattern recognition method who can simulate thinking mode of mankind to deal with problems and has dominant superiority to settle nonlinear problems. In this study , the signals come from gas sensor have nonlinear relationship with anticipant results. The appropriate choice of artificial neural network of BP type was made to resolve the problems effectively.The beef as sample in this experience came from Changchun HaoYue group and was put in a test-box that is conditioned similar to indoor environment(temperature: 22℃, humidity:75%, tested per 24hours). Firstly, forty samples were selected to train artificial neural network. And disciplined biomimetic electronic nose could distinguish days of beef storage at the same conditions. Another forty samples were used to check the validity of results tested by biomimetic electronic nose system.Another focus of this study is a comparison of the effects on the system performance of the array sensor be made up of different gas sensor. According to the test data of gas sensor, with responsive gas sensors replace the sensor insensitive response, and test performance differences by the experimental.It has proved by the experimental that the bionic nose flow system is simple and effective. The performance of the bionic electronic nose system would be changed if the gas sensor is replacement one or more. Thus the different electric nose system could be constructed for different samples, purpose or precision. According the test results, the best bionic electric nose to evaluate the freshness of beef was contracted. The bionic electronic nose system was inspected after training, the training sample-return rate could reach 100 percent and the judge correct rate of new samples around 90 percent.